Tech • AI • Robotics • Game

VIDEO
ENFR

New Elon Musk's Tesla Bot Optimus Gen 3 Capabilities Leaked

9/10
TeslaTESLA CAR WORLDOctober 17, 2026 at 11:24 AM28:07
Audio player

TL;DR

Tesla is reportedly preparing major advances for both its Optimus Gen 3 humanoid robot and its Cybercab production system, aiming to scale robotics and vehicle manufacturing through new hardware, AI and factory design.

KEY POINTS

Optimus shifts toward a more human-like design

Optimus Gen 3 is described as far more refined than earlier versions, with exposed joints, wiring and drive mechanisms covered by a smoother shell. The hips, knees and ankles appear redesigned to resemble human proportions more closely, while darker exterior components replace some of the more industrial-looking parts used before. The changes suggest a machine intended for long operating hours in factories and homes, where snagging, dust and wear matter as much as appearance.

The hands are central to the upgrade

The most important hardware change appears to be the hands, which are expected to reach 22 degrees of freedom, up from about 11 on Gen 2 and roughly 20 on Optimus 2.5. That would put the robot closer to the 27 degrees of freedom in a human hand and expand its ability to grip, rotate and manipulate objects. The design may also shift more drive mechanisms into the forearms, potentially reducing hand weight and improving precision.

New tactile sensing could broaden practical use

A new 3D tactile touch sensor system is reportedly in development, with sensing density in the hands rising sharply. Force sensing at fingertips and across the palms could let the robot adjust grip pressure in real time, allowing it to handle fragile items like eggs or glasses while still moving heavier objects. Target performance cited for Gen 3 includes lifting about 40 pounds or just over 18 kilograms, a sign Tesla is trying to combine warehouse strength with kitchen-level dexterity.

Onboard computing is expected to rise sharply

A future AI5 chip is expected to deliver about 8 times the computing power of AI4, along with more than 9 times the memory capacity and roughly 5 times the memory bandwidth. That matters because a humanoid robot must simultaneously perceive objects, track people, maintain balance, control joints and interpret spoken instructions. AI5 is reportedly aimed at mass production in 2027, so current units may still rely on earlier hardware.

Battery life may improve without a larger pack

Optimus Gen 3 is expected to keep a battery near 2.3 kilowatt-hours while switching to 4680 cells for better energy density. Combined with lighter structure, more efficient actuators and lower-power computing, operating time is projected to improve by about 40%. At an estimated 125 pounds or 57 kilograms, the robot could run for 6 to 8 hours on routine work, or about 5 hours under heavier loads, before needing 1 to 2 hours of charging.

Tesla appears to be scaling internal robot deployment first

Production in Fremont is reportedly ramping to hundreds of robots per week, with a possible path to 1,000 per week. The near-term goal appears to be internal deployment across Tesla facilities, where large fleets can gather data on manipulation, mobility, balance and failure modes. Discussion around 10,000 to 30,000 robots for testing and self-play points to a strategy of using real-world repetition to improve reliability before broad external sales.

Price and reliability remain major open questions

A long-term target price for Optimus has been framed around $20,000 to $30,000, but early commercial units may land closer to $60,000 to $80,000 because of complex hands, sensors and computing hardware. Tesla has also acknowledged that early robots will make mistakes, including falls or misunderstood commands. That issue is especially sensitive if the machine is meant for household work, child care or elder care, where safety standards would need to be far higher than for staged demonstrations.

Cybercab production is built around parallel assembly

Tesla is also redesigning vehicle manufacturing for Cybercab through an unboxed process. Instead of building a car in one long sequence, the factory assembles major modules such as the front, rear, side structures, roof and structural battery pack in parallel. Those modules are then joined in a final operation, reducing line length by about half, cutting floor space by roughly 40% and lowering labor needs by around 30%.

Gigacastings and molded body panels cut time and complexity

Cybercab uses large front and rear castings made on a proposed 50,000-ton Gigapress, far beyond the 6,200-ton and 9,000-ton presses used elsewhere. Tesla also pairs those castings with colored reaction injection molded exterior panels, which could eliminate the need for a conventional paint shop. That would remove one of the largest and most energy-intensive sections of a car plant while also reducing VOC emissions and simplifying panel replacement.

A 10-second manufacturing target underpins the strategy

Tesla says a Model Y currently takes about 34 seconds per vehicle at line rate, while the new system could allow roughly three to four Cybercabs in that same span. The drive unit line is also designed for a cycle time of 10 seconds and uses a rare earth-free motor intended to reduce dependence on materials largely sourced from China. If achieved at scale, the approach would represent one of the most aggressive pushes yet to shrink cost, factory footprint and production time in EV manufacturing.

CONCLUSION

Tesla is pursuing two linked bets at once: humanoid robots that can perform useful physical work and factories redesigned for extreme throughput. Whether those ambitions translate into reliable products will depend less on concept demos than on safe operation, durable hardware and sustained high-volume manufacturing.

Ask a question

More from Tesla